Note: I’m doing some final prep work for my SMX presentation tomorrow. Well, actually I’m sitting here in a daze after a good Swedish beer, trying to stay awake until 9 PM local time. But I’m rehearsing, too, and I often get my thoughts straight by writing about it, first. If you have a moment, please review my notes, and let me know if you’ve got any comments/additions. I only have 10 minutes to talk – the rest is a panel discussion. Thanks!
Measuring Pay Per Click Keyword Effectiveness
Analytics improvements in the last few years have made pay per click marketing quite a bit easier. If you have a clear conversion goal, like a sale, a lead, or a download.
What if you lack a conversion goal? Publisher sites, blogs and corporate branding sites focus instead on getting attention. That makes measuring the effectiveness of PPC keywords very difficult. Have you ever tried to tell your boss or client, â€œSure, it’s OK that we spent $25,000 on the word ‘cloth’ last month. It’s a great keyword!â€
It’s not pretty.
Tracking pageviews doesn’t necessarily work, either: Not all pageviews are alike. That’s true for visits, too. So we continue driving with our eyes closed. It works for a while, but a crash is inevitable.
What we need is a consistent solution for PPC waste management. You can get there if you apply two rules:
- Be consistently inaccurate
- Ignore nothing
Keyword Analytics Method 1: Eyeballs
With a little effort, you can measure keyword effectiveness by looking at your analytics reports:
All decent analytics software will show you the keywords visitors searched when they found your site. They’ll also split paid and unpaid search. The above report shows paid search only. At first glance, ‘conversational marketing’ is the winner. 26 visits! 2.38 page views per visit! Whoo-hoo! Spend more money!
But wait a second. Average time on site is only 53 seconds. They can’t be reading anything in that short a time. In this case, more pageviews isn’t better.
Truth is, none of these keywords are very good. Even ‘conversation marketing’, with its healthy 5 page views per visit, has an average time on site of only 1:24. That’s about enough time to glance at a page and move on.
The more analytically inclined will say ‘Wait, this is a small group, and Google’s not that accurate, and what about people who are refusing browser cookies…’.
All good points, but you have to assume some level of inaccuracy. It’s consistency that matters. If Google Analytics is always off by 10%, we can keep that in mind.
That’s rule #1: Be consistently inaccurate. You don’t need 100% accuracy. You need consistency, so that you can compare keywords to each other. If you don’t have a clear conversion goal, then relative keyword performance is almost as good as absolute keyword performance.
Rule #2 helped here, too: Ignore nothing. If I’d stopped at visits and pageviews, I’d be out buying myself another beer. By also including time on site, I quickly realized something was amiss. Don’t favor one metric over another. Ignore nothing.
Keyword Analytics Method 2: Formula
One of my colleagues, Matthew, has developed an actual formula for comparing keywords without conversions. I won’t even try to type it here. It gives me a headache just looking at it. Luckily, he was kind enough to place the formula in a handy Excel spreadsheet.
You’ll need 5 keyword metrics to make this work:
- Average visit count, per keyword: That’s the number of times a visitor returns to your site after finding you using that keyword;
- Total visits generated by that keyword;
- Average visit time, in seconds, for each keyword;
- Average page views per visit from that keyword;
- Target page time: The idea length, in seconds, of a pageview on your site. If you’re a video site, then target page time might be 600 seconds. If you’re a blog, it could be as little as 60 seconds. This number lets you adjust the formula for your site’s unique needs.
Plug those numbers in, and you’ll see a quality score in the right-hand column:
The higher the score, the better the keyword.
Don’t touch the ‘weight’ column. It adjusts for the fact that, on most sites, the second and third return visit is a far bigger ‘win’ than the 10th or 11th repeat visit.
This formula works, not because it’s precise, but because it’s consistently inaccurate, and because it ignores nothing (well, not nothing, but it ignores very little). It follows the two rules.
You can try the spreadsheet out for yourself: Download it here.
Note that this formula is a work in progress. If you have improvements, please, by all means, make them! Just make sure you let us know, too.